The Monocular-vision-based Measurement Method Of Geometric Errors Of Machine Tool Rotary Axis

CNC machine tools are the key means for modern production to be widely used in various manufacturing fields.As the demand for high-performance equipment manufacturing rises,the higher standards are set for the quality of machining.Multi-axis CNC machine tools are the important processing method for parts with complicated profiles,and the machining accuracy directly affects the dimensional accuracy and mechanical property of parts.However,the machine tool is affected by various error sources,leading to the decrease of accuracy,and the rotary axis geometric errors are the one of the main reasons for inaccurate machining.To improve the machining accuracy of the machine tools,the measurement,identification and compensation of the geometric errors for the rotary axis are of great significance.In order to meet the requirements of error separation and identification,the measurement procedure of the existing equipments may have problems such as cumbersome measurement processes and complex mathematical models with corresponding algorithms or algorithms due to the equipment structure restrictions,the limited measurement dimensions or sensitive direction constraints.Nevertheless,machine vision combined with catadioptric imaging technology has the ability of quick detection and integration of installation and measurement.Meanwhile,in view of the advantages of inexpensive cost,good integration and three-dimensional non-contact measurement,it has the great potential in the error detection of machine tool rotary axis.Therefore,a monocular-stereo-vision-based machine tool rotary axis geometric error measurement system and method is proposed in this paper.The primary research contents are as follows:(1)By analyzing the machine tool structure to obtain the error detection requirements,a rotary axis geometric error measurement method based on the monocular stereo vision is proposed.Firstly,according to the kinematic analysis of the machine tool motion,the spatial action mechanism of the rotary axis geometric errors on the space points is discussed,and the geometric error model of the rotary axis is established by compassion.Secondly,a monocular stereo vision imaging model is established based on the law of reflection.Meanwhile,the mapping relationship between spatial points and corresponding pixel points in the monocular stereo vision system is described.Finally,the monocular stereo vision overall measurement scheme for the rotary axis geometric errors is established,and the functions of each part of the hardware system and the overall identification process are explained.(2)In view that the cost and measurement dimension cannot be taken into consideration,and field of view and measurement accuracy limit each other in the vision measurement method,a monocular stereo vision measurement system for the rotary axis geometric errors is designed and built.Based on the flexible and adjustable monocular stereo vision acquisition system,the corresponding optical path simulation analysis method is proposed,and the relationship between system structure parameters is analyzed.For the requirement to accurately characterize the global motion positions of the rotary axis,a feature target with high uniformity,high precision and high signal-to-noise ratio is designed using the idea of error amplification.In addition,to meet the requirement for rapid adjustment of system parameters,the monocularstereo-vision accuracy analysis model is derived through simulation,realizing the quantitative evaluation and optimal selection of key system indicators.(3)In order to accurately identify the rotary axis geometric errors of the machine tool using pixel coordinates,the camera calibration,target feature extraction,reference transformation and error separation and identification methods are studied with the established monocular stereo vision system.Firstly,a monocular stereo vision high-precision calibration method considering the image scale and depth of field is proposed to realize the efficient processing of sequence images.Finally,an error separation and identification algorithm based on the principle of multipoint positioning and a transformation method from vision coordinate system to machine tool coordinate system are proposed.According to the above techniques,the measurement experiment and accuracy verification tests are developed.The results show that the average linear errors and angle errors measured by monocular stereo vision system are 0.71 ?m/mm,0.26 ?m/mm and 0.56 ?m/mm,and the average angle errors are respectively 0.013°,0.014° and 0.028°.Meanwhile,the system uncertainty 7.84 ?m is less than 1/3 of the measurement average spatial error,verifying the effectiveness and robustness of the proposed method in this paper.